Alan Cox wrote:
> 
> > > UDI is not a standard and seem to me some poor sci-fi and will never
> > > happen in my systems.
> >
> > I just browse http://www.projectudi.org/ and find out that Intel already
> > porting UDI  in Linux as proof of concept.
> 
> That doesn't mean anyone in the Linux community considers UDI a serious
> exercise.

And no one I know considers it efficient.  Nor would I consider it SMP
"friendly" with the granularity of its locking (not that the current linux
SCSI locks don't suck as well, but I wouldn't use UDI as an example of doing
things right here).

> > 1. support by major vendors like Sun, HP, IBM and Compaq (which buyout DEC and
> > abandon CAM).
> 
> Vendors and clues don't go hand in hand. FreeBSD has more clues than all the
> vendors put together and a lot less marketing reasons to choose stuff their
> "friends" say they should. FreeBSD SCSI is rather nicely done.

For certain classes of controller (basically, if it has a firmware that
wants/needs to do things itself on its own timeframe, etc. then I've heard
some of those authors grumbling about the amount of controler moved away from
the low level driver to the CAM layer).

> > 2. Intel is porting UDI for Linux now!
> 
> And people are porting cobol compilers. Neither is a good idea.
> 
> > 3. UDI support SCSI-3 (including Fibre Channel)...Do CAM support that?
> 
> Of course it can, and ultra-3 LVD and all the rest. It also does it without
> throwing your latency and performance out of the window.

Not to mention that UDI isn't intended to be a complete SCSI stack (it doesn't
interface directly with disks or tapes or anything else).  UDI is a driver
frame work, nothing more.  For the higher layer services such as disk and tape
drivers you still have to do your own work.  If you're going to be doing that
work, there is no reason to do it as UDI instead of something native and of
far superior performance/flexibility/functionality.

-- 
  Doug Ledford   <[EMAIL PROTECTED]>
   Opinions expressed are my own, but
      they should be everybody's.

-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to [EMAIL PROTECTED]

Reply via email to